Game Mechanics: AC may not be working correctly

[Retti_]

My cleric has highest ac on server

[Holey]

as tested, backstabbing only on warriors i was doing average of less than 200 i might get lucky and pull off a 400+ on an sk.

[Aeolwind]

IIRC AC didn't do much on live either. It was a byproduct of pumping HP/STA/Relevant stat. It wasn't until Velious that it actually started to matter. AOW & Lord Vyemm made it pretty important.

Most of the mobs in Velious had sick, disgusting AC's. The '1' spam on Vulak is rather humorous from the logs I have.

KEEP WORKING ON THIS however. It is important to get the DB/DI working correctly with AC/Defense in general. It'll make it easier for me to kill you then =D.

[Splorf22]

Aeolwind I'm told there is a #AC command for gms. How about you play with it and give us the results? I only did 150 hits each, so its not impossible I was just extremely unlucky.

[Aeolwind]

Originally Posted by Splorf22 [You must be logged in to view images. Log in or Register.]

Aeolwind I'm told there is a #AC command for gms. How about you play with it and give us the results? I only did 150 hits each, so its not impossible I was just extremely unlucky.

Give me a list of gear to have equipped, class, level, mob you want me to get pummeled by and the # the skills should be at. I'll let Nil decide what to do with the info after I gather it.

I'm not sure there is an AC Command honestly, never went looking. But using it would/could taint the results.

[Splorf22]

Originally Posted by Kahvok

Aggregate data from live servers at the time was taken to determine median-AC stats for each class. Parses were run against NPCs 3-4 levels lower, facing front. The characters had cleric AC and shaman agility buffs and faced the NPC. The results of the parse were consistent with statistical analysis of the formulas in code:

Class War Pal Mnk
Level 51 51 51
Raw Item AC 184 181 107
Agility 157 144 169
Dodge 3.4% 3.1% 4.4%
Block 0 % 0% 10.2%
Riposte 4.4% 3.9% 4.1%
Parry 5.2% 4.6% 0%
Skill Evasion 12.9% 11.5% 18.7%
Hit Rate 61.2% 61.3% 58.2%
Avg Hit 72.6 72.9 74.6
% Hits for Max 10.2% 10.5% 11.5%
Avg Dmg / Round 59.7 61.1 54.5
DPS 28.2 28.8 25.7


Class War Pal Mnk
Level 60 60 60
Raw Item AC 296 281 163
Agility 177 152 187
Dodge 4.3% 3.9% 4.9%
Block 0 % 0% 11.4%
Riposte 4.8% 4.3% 4.5%
Parry 5.8% 5.2% 0%
Skill Evasion 14.9% 13.4% 20.8%
Hit Rate 59.4% 59.7% 59.3%
Avg Hit 107.3 109.9 113.6
% Hits for Max 10.4% 11.7% 13.6%
Avg Dmg / Round 87.4 91.7 86.1
DPS 50.8 53.3 50


The problem was that the average plate-equipped warriors and knights had barely any lead on monks in mitigation, due to the monk bonus, but the monk still had the lead in evasion. Contrary to popular belief, this is what prompted the nerf to monk mitigation, NOT high-end monks being rampage tanks.

The changes had little effect on average level 51 warriors and knights, but since the average level 51 monk was over the new nerfed AC cap, it increased their average damage taken per hit and increased the percent chance of max hits (in the above example) to 13%. Monks who had better than this median AC were hit harder by the nerf since it lowered their effective AC even more. Level 60 monks with exceptionally high item AC (Ssra+) weren't hit quite as hard because the uncapping of item AC gave them more returns on AC over the class cap. The median level 60 changes looked like this (evasion, of course, remained the same):

Class War Pal Mnk
Avg Hit 106 108.9 121.3
% Hits for Max 9.8% 11.2% 18.4%
Avg Dmg 86.4 90.9 91.9
DPS 50.2 52.8 53.4

It should look like this. 51 and 60 warriors/monks, items to that that amount of AC, etc.

Also my understanding is the #AC command does not set your AC to X (which would be plausible) but instead simulates multiple hits against a given AC. However one of the GMs assured me that AC was indeed working and based his opinion on this command, so <shrug>

[Alecta]

Disclaimer: I cant step through my debugger right now, it's late, and I am out of town for the week, so this information might be wrong.

The AC system isn't very clean, but from what I can tell:

• The shield AC isnt handled any differently from normal AC
• Soft caps arent in place (?)

These are both due to the fact that the "NPC hitting a target" code uses the AC variable (which is what is displayed client side) and doesnt use the RealAC variable (which is what takes into account soft caps, shield AC, etc).

Also, I'd ignore the #ac_test command since that simulates "Player attacking something" and not "NPC attacking something" - and those follow different code paths.

NPC::ProcessAttack -> Mob::MeleeMitigation -> defender->GetAC()
vs
Client::ProcessAttack -> Client::ThrowATKvAC_D20I -> defender->GetRealAC()

Off topic: I am curious if the Monk AC bonus is working properly all the time. It seems that due to the order the bonuses are calculated, it might not be.

Anyway, will post more when I have more time to dig into this.

[Splorf22]

Originally Posted by Alecta [You must be logged in to view images. Log in or Register.]

• The shield AC isnt handled any differently from normal AC
• Soft caps arent in place (?)

I'm guessing that this code implements the post from Treats I linked in the OP. According to him both the soft cap and the shield AC thing are from the Luclin rework.

[koros]

Parses seem to indicate softcaps aren't in place, so that makes sense. Can you share the code of the methods utilized so we can see if they match up with the softcap info from the original devs? It's possible people just have a lot more ac now than they had on live and are noticing it to a much greater degree.

[Alecta]

Originally Posted by koros [You must be logged in to view images. Log in or Register.]

Can you share the code of the methods utilized so we can see if they match up with the softcap info from the original devs?

Nilbog said he was fine with it, waiting on permission from Rogean.

Again, I'm pretty new here and am really just supposed to work on PvP stuff, so I am not comfortable posting other peoples contributions to the code base.